"We talk about the risks, what about the gains?" Large language models and learning in mathematics, statistics, and computing

Abstract

Large language models are increasingly present in higher education worldwide, yet debates in many institutions continue to focus mainly on concerns about academic integrity, misuse, and possible adverse effects on student learning. This paper presents a conceptual and argumentative discussion based on an extensive review of recent international and African literature on mathematics, statistics, and computational learning.

The review explains how large language models can strengthen quantitative learning when supported by responsible institutional policy, ethical guidance, and sound pedagogical practice. Evidence from the literature indicates that these systems can support stepwise mathematical reasoning, improve statistical understanding, strengthen programming competence, enhance academic writing and reporting, and increase student motivation, particularly where timely human feedback is limited.

The paper argues that successful integration depends on AI literacy, effective prompting skills, guided classroom use, verification practices, and continued human oversight. It further discusses the need for curriculum reform, assessment redesign, and institutional readiness within African higher education while recognising challenges relating to access, equity, and digital infrastructure.

The paper concludes that large language models should not be viewed solely as threats to higher education. When introduced responsibly, they can improve learning quality, widen educational participation, and strengthen competence in mathematics, statistics, computing, and other quantitative disciplines that are important for national development across Nigeria and the wider African region.

Keywords

Artificial intelligence, African higher education, computational learning, large language models, mathematics education, quantitative education, statistics education

Citation